Epilepsy is one of the most common neurological syndromes, affecting an estimated 3 million people in the United States. In one-third of these patients, seizures cannot be controlled despite maximal medication management. The complexity of the neuronal network dynamics that define the epileptogenic cortex and drive seizure initiation and spread makes understanding and treating epilepsy a unique challenge. In this proposal, an interdisciplinary research team will address this challenge. The assembled researchers integrate clinical expertise and data recording capabilities with sophisticated network analysis and statistical modeling techniques. Utilizing invasive brain voltage recordings, dynamic functional networks will be inferred from a population of patients during spontaneous seizures. To characterize these dynamic networks, new data analysis and statistical modeling techniques tailored to address the unique challenges of the clinical human data will be developed. These techniques will be applied to understand the sudden, explosive emergence of well-connected subsets of nodes (a.k.a., communities) in the noisy, real-world environment of human cortical seizure dynamics. Understanding the rapid network organization at seizure onset and termination will inspire new treatment strategies for epilepsy, and motivate developments and applications in the emerging theoretical research field of explosive percolation. The proposed research will advance scientific knowledge and understanding in three ways. First, the development and application of novel dynamic network analysis techniques to clinical seizure data will provide a deeper understanding of human epilepsy and the network interactions that underlie seizure initiation and termination. Second, the proposed research requires new tools to characterize and track community structure in noisy, dynamic networks. Development of these tools will help to address open questions and unexplored directions in the study of transient and recurrent community patterns emergent in dynamic networks. All dynamic network analysis tools developed in this proposal will be made freely available for other researchers to apply and develop. Third, by utilizing complex neurophysiological data, the proposed research will ground the field of explosive percolation in noisy real-world phenomena, and motivate new developments and applications critical to this emerging science. There are three broader impacts of the proposed research. First, the dynamic network analysis and statistical modeling of human seizure data will provide new approaches to improve patient care of medically refractory epilepsy. In particular, through prospective and retrospective studies, the dynamic network analysis and modeling techniques will be applied to identify principled surgical targets, and predict which patients will - and will not - benefit from surgery. Second, the dynamic community detection tools and statistical models developed will have general applicability across many domains of science. These tools can be applied broadly within systems neuroscience - to elucidate brain dynamics underlying healthy brain function and present in pathology - and in many other scientific fields (e.g., cell biology, ecology, social sciences, distributed computing, to name a few) in which dynamic networks appear. Third, the proposed research will provide unique training opportunities for graduate students in translational neuroscience, with a specific emphasis on clinical data, network inference and dynamical network analysis, and statistical modeling. These trainees will develop unique interdisciplinary skills in clinical, statistical, and computational neuroscience.